Solubility of SLs in various solvents

The motivation for updating our SL component analysis method using HPLC was the presence of undissolved SL components in the analysis sample obtained using the conventional method. Therefore, we evaluated the solubility of the SL components in common solvents. Our results are summarized in Table 1. The well-known LSL and ASL showed opposite solubility in polar and nonpolar solvents, as expected. SLGs, the third SL type identified in our previous study, also showed differences in solubility that depended on the specific structure. Additionally, although LSL alone is insoluble in water, S. bombicola products that were mixtures of LSL and ASL were easily dispersed in water. Taken together, the results of the above solubility test showed that LSL, ASL, and mixtures thereof dissolved completely in methanol, thus showing great promise for use as an HPLC eluent.

Table 1 Solubility of sophorolipids and related compounds in various solventsaComponent analysis of S. bombicola products using reversed-phase HPLC and LC–MS

Based on the above results, we attempted to analyze all SL components in S. bombicola products using reversed-phase HPLC in a methanol solvent system. Figure 2 shows an HPLC chart of S. bombicola products obtained from rapeseed oil by jar fermentation, as described in the experimental section. The peaks were assigned to ASL (group A in Fig. 2) or LSL (group B in Fig. 2) by analyzing the purified ASL and LSL samples. We previously reported that the three large peaks (peaks C–E in Fig. 2), eluted in the latter half, correspond to SL di- and tri-glycerides (Kobayashi et al. 2023). We further focused on three peaks containing unknown compounds (peaks X at retention time [r.t.] around 4.3 min, Y at r.t. around 9–10 min, and Z at r.t. around 19–20 min in Fig. 2) by comparing them with our LC–MS analysis results (Table 2).

Fig. 2

Charged aerosol detector high-performance liquid chromatography (CAD-HPLC) analysis of the collected microbial products of S. bombicola by jar fermentation from rapeseed oil using a C18 silica gel column and methanol/water eluent

Table 2 Peak list on HPLC chart of a S. bombicola product

Table 2 shows that the mass of compound X [M-H]– (m/z) was 1,029.5 in fraction No.1; this is different from any of the known values for LSL, ASL, or SL di- and tri-glycerides. The structure estimated from the molecular weight value of compound X is an SL derivative in which ASL is esterified to another disaccharide (probably sophorose). This compound is most likely “bora-form SL”, synthesized recently by culturing genetically modified microorganisms (Van Bogaert et al. 2016; Van Renterghem et al. 2018).

Other newly identified compounds included compound Y [M-H]– (m/z), with a mass of 779.4 in fraction No.6, and compound Z [M-H]– (m/z) with a mass of 985.6 in fraction No.16 (Table 2). Based on the structures estimated from the molecular weight values of compounds Y and Z, Y is an SL mono-glyceride, in which one ASL is esterified to glycerol, whereas Z is an SL derivative in which another fatty acid is esterified to ASL. To our knowledge, compound Z is a newly identified SL derivative; however, further analysis is required to determine its structure as this compound has many possible isomers.

Isolation and identification of compounds X–Z

The three compounds X–Z were isolated by careful fractionation using C18 silica gel column chromatography. Compounds X and Y were trace components in the S. bombicola product. They could not be isolated by normal-phase silica gel column chromatography because they could not be separated from ASL. Therefore, we first separated LSL from the S. bombicola product using normal-phase silica gel column chromatography and recovered the remaining SL components, including ASL and other SL derivatives. Next, the remaining SL components were fractionated sequentially and precisely using reversed-phase C18 silica gel column chromatography. The results of LC–MS and NMR analyses of the isolated and purified compounds X and Y are shown in the Supplementary Information (Figs. S1–S9 and Table S1, S2). As expected, the estimated X and Y main components were bola-form SL and SL mono-glyceride, respectively (Fig. 3).

Fig. 3

Putative chemical structures of the main components of three compounds newly found in S. bombicola products by CAD-HPLC and liquid chromatography- mass spectrometry (LC–MS) analysis. (a) Compound X (bola-form sophorolipid), (b) compound Y (sophorolipid mono-glyceride), and (c) compound Z

In contrast, the peak of compound Z completely overlapped that of SL tri-glyceride (SLG-B) on the reversed-phase HPLC chart (Fig. 2); however, these compounds can be separated. First, compound Z including LSL was separated from the S. bombicola product using normal-phase silica gel column chromatography. Compound Z was then separated from LSL and isolated using reversed-phase C18 silica gel column chromatography (Fig. S10 in the Supplementary Information).

Purified compound Z dissolved in chloroform, whereas the SL derivatives other than LSL did not. Detailed structural analysis of compound Z dissolved in deuterated chloroform was performed by 1D and 2D NMR analysis. The 1H NMR spectrum of compound Z (Fig. 4) was similar to that of the well-known ASL. Importantly, H-4'' of sophorose did not shift to a lower field, confirming that the hydroxyl group at the 4'' position was not esterified and neither lactone nor other esters formed. Similarly, it was confirmed that no other hydroxyl groups of sophorose (3', 4', 2'', 3'', and 4'' positions) had any ester bonds except for acetylation at 6' and 6'' positions. Focusing on the two triplet peaks from 2.2 to 2.4 ppm, which are not found in conventional ASLs, these peaks originate from the protons of the methylene group next to the carbonyl (i.e., carboxyl) groups (-CH2CO-). In particular, based on the difference in the positions of the two peaks, it is expected that one is a free carboxylic acid and the other is an ester.

Fig. 4

The 400-MHz proton nuclear magnetic resonance (1H-NMR) spectrum of compound Z, isolated from S. bombicola

It is important to note the other peaks at 4.06 and 4.89 ppm, which are not found in well-known ASLs. The multiplet peak at 4.89 ppm correlates with the doublet peak at 1.19 ppm (CH3CHOH-) and the broad peak at 1.5 ppm (-CH2-) in 1H-1H COSY (Fig. S5, Supplementary Information). This peak is associated with the proton at the ω-1 position, which has the same structure as the ω-1 position hydroxy fatty acid in the SL molecule. However, in conventional SLs, the hydroxyl group of the hydroxy fatty acid is connected to sophorose through an ether bond, whereas the proton at the ω-1 position appears at 3.78 ppm. The peak at 4.89 ppm was shifted toward the lower magnetic field; this implies that the hydroxy group at the ω-1 position is esterified. Based on these results, the peak at 4.89 ppm is derived from the proton of CH3CH(OCO-)-. Similarly, the triplet peak at 4.06 ppm is correlated with the broad peak at 1.61 ppm (-CH2-) in 1H-1H COSY (Fig. S11), derived from the proton at the ω position of ω hydroxy fatty acid; this hydroxyl group at the ω position is esterified (-CH2OCO-). Further, our HMBC analysis results (Fig. 5) confirmed that the carbon of the carboxyl group of the first fatty acid at 173.8 ppm (C = O of FA1) is correlated with the protons at 2.27 ppm (CH2C = O of FA1) and at 4.06 ppm (-CH2OCO- of the ω hydroxy FA2) or 4.89 ppm (CH3CH(OCO-)- of the ω-1 hydroxy FA2); these are the protons at the carbon to which the hydroxyl group of the second hydroxy fatty acid (FA2) is attached. In other words, our results showed that the carboxylic acid of ASL (FA1) forms an ester bond with the hydroxyl group of FA2. On the other hand, the carbon of the carboxyl group of the second fatty acid at 178.5 ppm (C = O of FA2) only correlates with the protons at 2.33 ppm (CH2C = O of FA2). Thus, the FA2 is a free carboxylic acid. Based on the above results, compound Z, which has [M-H]– (m/z) of 985.6, the same carbohydrate structure as ASL, and two fatty acids (one of which is an ester and one of which is a free carboxylic acid) is estimated to have the structure shown in Fig. 3(c). The results of 13C NMR and HSQC (Fig. S12) analysis confirmed these findings (Table 3).

Fig. 5

Partial heteronuclear multiple bond coherence (HMBC) spectrum of compound Z, isolated from S. bombicola, F1 axis: Carbon 13 NMR (13C-NMR) spectrum ranging from 165 to 185 ppm; F2 axis: 1H-NMR spectrum ranging from 0.6 to 6.0 ppm

Table 3 NMR data for the compound Z (chloroform-d, 400 MHz)Properties of compound Z

As mentioned above, compound Z has a completely unique chemical structure, in which two long-chain fatty acids are connected in series to a disaccharide; this feature is rare in common surfactants. We attempted to demonstrate the surface activity of this compound by measuring the surface tension of its aqueous solutions. When compound Z was dispersed in water, it became cloudy and formed an aggregate at lower concentrations (< 1 g/L), unlike ASL, which is highly soluble in water. Unexpectedly, we observed that even when compound Z was dispersed in water at a higher concentration (~ 20 g/L), the surface tension decreased only to 61.1 mN/m. Furthermore, in highly concentrated aqueous solutions (> 20 g/L), the insoluble part precipitated, and measurements could not be performed. From this result, compound Z has very low surface activity.

SLs are well known for excellent antimicrobial activity. We examined the antimicrobial activity of the newly found compound Z against gram-positive and gram-negative bacteria. Three bacteria (B. subtilis, S. aureus, and E. coli) were inoculated in liquid media containing SL derivatives, and the bacterial growth was observed. Compound Z did not exhibit antimicrobial activity against any bacteria in the concentration range tested (< 1,024 µg/mL), whereas LSL showed activity against gram-positive bacteria.

Investigation of changes in component ratios in S. bombicola products

As demonstrated, with the appropriate solvent, it is possible to analyze all components of SLs using HPLC and how the component composition changes over time in the cultured products of S. bombicola. Figure 6a–c show the changes in the components of S. bombicola products obtained from rapeseed oil after jar fermentation for 3 days. The culture solution was diluted directly with 70% methanol, filtered, and subjected to HPLC analysis. The quantification of SL derivatives and the culture profile are shown in Fig. 6g and Fig. S13 in the Supplementary Information. LSL (r.t. 11.5–16.8 min) was produced as the main component from the early stage of culture (~ 24 h). In contrast, the content ratio of the third component, the SLGs, (three peaks around r.t. 16.8–24 min) increased from the middle to late stages of culture (1.0 g/L at 24 h, 28.3 g/L at 48 h, and 32.4 g/L at 72 h), indicating that their accumulation in the culture solution was progressing. Furthermore, comparing the results after 48 and 72 h (Fig. 6b, c, and g), SLG-A–C tended to be converted to SLG-B (SL tri-glyceride: Fig. 1d, the peak at r.t. 19.3–20.5 min) (SLG-A:B:C = 7.0:16.2:5.1 [g/L] at 48 h, and 9.1:20.1:3.2 [g/L] at 72 h).

Fig. 6

Component analysis of the culture medium of S. bombicola by CAD-HPLC using a C18 silica gel column and methanol/water eluent. Using rapeseed oil as a carbon resource at a) 24, b) 48, and c) 72 h, or using rice bran oil at d) 24, e) 48, and f) 72 h after the start of fermentation, and g) Quantification of the SL derivatives

We changed the type of vegetable oil and used rice bran oil to track similar changes in the component ratio of S. bombicola products (Fig. 6d-f). The trend of the overall SL components was confirmed to be similar to that when rapeseed oil was used. That is, LSL was the main component from the early stage of culture throughout the entire period; SLGs increased from the middle to late stages of culture (1.9 g/L at 24 h, 26.5 g/L at 48 h, and 30.8 g/L at 72 h); SLG-A–C tended to be converted to SLG-B (SLG-A:B:C = 6.7:11.9:8.0 [g/L] at 48 h, and 8.6:16.7:5.6 [g/L] at 72 h). When rapeseed oil was used, the fatty acids constituting ASL and LSL were rich in C18:1 (r.t. 8–9 min and 15.2 min). However, when rice bran oil was used, the ratio of C18:2 and C16:0 (r.t. 6–7 min and 14 min) increased significantly.